Validate the output of the onnxruntime backend is the same as the output generated by torch. Args: model (nn.Module | function): the function of model or model to be verified. feats (tuple(list(torch.Tensor)) | list(torch.Tensor) | torch.Tensor): the
(model, feats, onnx_io='tmp.onnx')
source not stored for this graph (policy: none)